INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In 1982, a new pathogen causing hemorrhagic colitis emerged that later became know as Escherichia coli O157:H7. Since then, tremendous efforts have been put forth to identify and characterize this enteropathogen. About 20,000 cases per year of E. coli O157:H7-caused hemorrhagic colitis are found, and this number is increasing. Approximately 5% of the patients develop more serious health problems, such as hemolytic anemia, kidney failure, and thrombocytopenia (27). The route of infection is usually fecal-oral transmission.

E. coli O157:H7 contamination is of concern to the food industry because of the pathogenicity of this organism and the increase in cases (31). It is found in the environment and is prevalent in domestic farm animals, having been isolated from calves, cattle, and sheep (7, 10, 45). Thus, potential cross-contamination at the farm or in commercial meat processing plants can lead to infections from foods (13, 23). Apple juice, apple cider, raw apples, milk, ground beef, radish sprouts, salami, tomatoes, and lettuce have been associated with outbreaks from food sources (2, 5, 9, 11, 19, 28, 35). Other documented outbreaks involve swimming pool and drinking water (30, 41).

Over the past several years, rapid detection methods have been developed for E. coli O157:H7, but all still require at least 6 h of the preenrichment step before the detection phase. Sorbitol MacConkey medium has been the medium of choice in isolating and identifying non-sorbitol-fermenting E. coli O157:H7, followed by additional testing to verify the identification (8, 17, 25). deBoer (12) summarized recent developments in isolation tools for use with solid media. Detection of E. coli is with membrane filtration followed by growth on selective agar containing chromigenic and fluorogenic substrates as an indicator of beta-D-glucuronidase activity; however, this test is not specific for E. coli O157. Further verification after these procedures requires serotyping of the isolates.

Enzyme-linked immunosorbent assays (ELISAs) for detection of E. coli O157:H7 were developed to meet the need for faster detection. ELISAs are performed after the preenrichment step and often require only minutes to visualize the results in a lateral flow device (11, 21, 23, 26, 35). Together with various enrichment methods and ELISA-based detection methods, the analysis time and sensitivity have improved, taking less than 24 h.

Several research groups have developed immunomagnetic separations for the detection of E. coli O157:H7. These methods still require overnight preenrichment followed by capture and concentration of the magnetic beads prior to detection using an ELISA (7, 22, 34, 36, 37, 44). The sensitivity ranges from 10 to 10² CFU/g of ground beef. These tests take more than 8 h to run after preenrichment.

Other research groups have focused on developing filters or other solid supports to capture and concentrate E. coli O157:H7 (11, 23, 29). ELISAs can be performed on these solid supports, and the sensitivity is between 0.1 and 1.3 cells/g of ground beef. However, filtration presents new cell collection problems in complex samples that may clog the filter, thereby limiting the sample volume that can be used for the test.

Numerous comparative studies of detection and identification methods available on the market have also been made (15, 20, 24, 33, 39, 40, 43). They all include preenrichment steps followed by a detection method. The sensitivity and lengths of the tests are similar. Thus, the focus has shifted toward developing methods that omit the preenrichment step to reduce analysis time. To try to meet the collection demand, Tortorello and Stewart (38) developed an antibody-direct epifluorescent filter technique. The sample is homogenized, treated with trypsin and Triton X-100, and concentrated onto a 0.2-µm-pore-size polycarbonate filter. The filter is subsequently stained and analyzed by epifluorescence microscopy. The sensitivity of this test is 16 CFU/g, and the test time was reduced to <1 h. Seo et al. (36) developed a test combining immunomagnetic bead separation with flow cytometry. In <1 h, 10³ to 10⁴ CFU/ml of ground beef was detected. Gehring et al. (14) developed an immunoligand assay. The sensitivity of 2.5 × 10⁴ cells/ml was reached in 30 min. Brewster and Mazenko (4) developed a filtration technique, which obtained a sensitivity of 5 × 10³ CFU/ml in 25 min. Abdel-Hamid et al. (1) developed the most sensitive method, requiring no preenrichment, by using a flow-through immunofiltration system. This combines flow-through, immunofiltration, and enzyme immunoassay techniques to achieve a sensitivity of 100 cells/ml in 30 min. However, clogging has caused this test to fail with complex food samples.

The need to develop faster and more sensitive methods is apparent. Our research group has focused on developing a biosensor that is more sensitive than 100 cells/ml. This ImmunoFlow system involves the use of existing ELISA protocols together with a new flow-through capture mechanism. We have developed a system which will identify small proteins, i.e., ovalbumin (OVA) and bovine serum albumin (BSA), Bacillus spores, and E. coli O157:H7. The sensitivity for Bacillus and E. coli O157:H7 is <10 total cells independent of the sample size. This new ImmunoFlow (patent awarded) method can easily be adapted for detection and identification of other pathogenic food-borne bacteria, such as Salmonella and Listeria (C. Beer, R. Koka, M. Hill, X. Wang, and B. Weimer, unpublished data).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Antibody characterization. Four clones of E. coli O157:H7 monoclonal antibodies (Ab) and one clone of E. coli K-12 monoclonal Ab were screened for cross-reactivity. All E. coli Ab were screened for cross-reactivity and ability to capture E. coli O157:H7 (ATCC 35150, 43888, 43889, 43894, and 43895) using sandwich ELISA procedures (16) (Table 1). Four anti-Bacillus spore monoclonal Ab were screened for cross-reactivity and ability to capture Bacillus globigii spores. All Ab used were immunoglobulin G (IgG) type affinity purified using an A/G column (Pierce Chemical, Rockford, Ill.) either by the manufacturer or by our laboratory according to the manufacturer's instructions.

Surface modification. The capture ability of antibodies attached to different spacers was investigated. Glass and ceramic surfaces were modified by the methods of Blake and Weimer (3) and Weimer et al. (42). Dextran (molecular weight, 37,500) (Sigma), polyethylene glycol-dicarboxylmethyl (PEG; molecular weight, 3,400) (Shearwater Polymers, Inc., Huntsville, Ala.), and polythreonine (molecular weight, 12,100) (Sigma) were used as spacers (Table 2).

Detection in static environment. Eight grams of Ab-modified beads was placed into a 50-ml centrifuge tube, and 10 ml of sample was added to the beads. Samples were incubated on a rocker for 1 h at 25°C. The samples were washed six times each with 50 ml of PBST (pH 5.8). Secondary Ab was added (total, 10¹² molecules of anti-E. coli O157:H7, 10¹³ molecules of anti-OVA, 10¹³ molecules of anti-BSA, and 10¹² molecules of anti-B. globigii) in 10 ml of PBST, and beads were again incubated for 1 h. Samples were washed six times with 50 ml of PBST (pH 5.8) and incubated with 10 ml of anti-IgG conjugated to HRP (Pierce) (IgG-HRP, 1 µg per 10 ml of PBST [pH 5.8]). After the last wash step, beads were added to 5 ml of 1-Step Turbo TMB-ELISA substrate (Pierce) and incubated in the dark for 20 min before an A₃₇₀ reading was taken using a Cary 100-Bio UV/visible light spectrophotometer (Varian, Sugar Land, Tex.). Water blanks were used as the control for the double-beam instrument.

Detection using ImmunoFlow. ImmunoFlow used a fluidized bed of beads, 8 g for the small unit and 250 g for the large unit, with Ab covalently bound. To generate flow, a vacuum pump was used. The reagents were evacuated from the bead cartridge through the top of the reactor at a constant rate of 0.4 liters/min (or 5 in. of Hg). As soon as all the liquid passed over the beads, the next reagent was allowed to flow through the reactor. This continued until all the reagents flowed across the beads. Just before the substrate (TMB) was added to the bead cartridge, the vacuum was turned off and the TMB was pulled into the reactor with a syringe. Once the TMB solution covered the beads, the cartridge was sealed and placed in the dark for 20 min. To measure the color development at A₃₇₀, 1 ml of the substrate was placed in a cuvette. Water blanks were used as a control for the double-beam spectrophotometer.

Stability of modified beads. Stability of anti-B. globigii spore modified beads was tested over a 5-week period. Two spacers, dextran and PEG, were used to attach the Ab to the surface of 3-mm-diameter glass beads at 10¹⁶ molecules/m². Two identical sets of experiments were run. One set of beads was stored in PBST (pH 7.2) containing 0.02% thimerosal at 8°C. The other set of beads was transferred into a stainless-steel module and placed in the river. The flow of the water ran straight through the module. River temperature and pH were recorded during the 5 weeks. Bead samples were taken once a week and tested for the ability to bind 10⁵ total B. globigii spores.

Confocal microscopy. B. globigii spores were labeled with Fura-indole C18 using a labeling kit from Molecular Probes (Eugene, Oreg.) according to the manufacturer's manual. Labeled spores showed a blue color under fluorescence (E_x, 481 nm; E_m, 694 nm). E. coli O157:H7 cells were stained with the LIVE/DEAD BacLight kit (Molecular Probes). Live bacteria showed a green color (E_x, 480 nm; E_m, 520 nm), and dead bacteria showed a red color (E_x, 550 nm; E_m, 580 nm) under fluorescence.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cross-reactivity of antibodies. All of the E. coli O157 Ab tested reacted to all strains of E. coli O157:H7. Escherichia coli K-12 cells did not cross-react with any of the E. coli O157 Ab but reacted to their own Ab. Because of this, the monoclonal mouse anti-E. coli K-12 Ab was used as a negative control in the experiments.

Static capture ability of modified beads. Figure 1 shows the standard curve obtained with anti-BSA-modified immunomagnetic beads and static detection. Ab-modified polystyrene beads, 10⁸ total beads, successfully captured BSA. Very small amounts (<1 ng) of BSA can be detected with these beads. The linear response of the signal to the increase in BSA was 99.7%, which makes this test very sensitive.

View larger version (8K): [in this window] [in a new window]   FIG. 1.   Standard curve of BSA capture with tosyl-activated poly-Thr-modified immunomagnetic beads. Standard errors of the means are masked by the symbols. The y axis indicates absorbance.

View larger version (8K): [in this window] [in a new window]   FIG. 2.   Standard curve of OVA capture using 3-mm-diameter PEG-modified glass beads. Error bars represent standard errors of the means.

Flow capture ability of modified beads. Figure 3 shows the capture of B. globigii spores at various flow rates in a continuous cycle. Figure 3A shows the capture efficiency in 0.25 M sodium phosphate buffer, pH 7.2, and Fig. 3B shows the efficiency in river water. Unmodified beads served as controls and were tested at the same time a sample was collected. Values for controls were subtracted from the sample signal, and the percent capture was calculated. Spores suspended in sodium phosphate buffer were not captured efficiently at the flow rates tested. Only 13% capture was seen within the first 45 min at 4 liters/min, and <10% was observed at lower flow rates. However, the spore count showed that most of the spores were captured at the end of the test period (data not shown). Spores spiked into river water were captured more efficiently. After 15 min, 70% of the spores were captured by the beads, but the capture efficiency dropped to <10% after 15 min.

View larger version (33K): [in this window] [in a new window]   FIG. 3.   Capture ability of anti-B. globigii-modified polythreonine ceramic beads in sodium phosphate buffer (pH 7.2) (A) and river water (B).

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Standard curve of capture of B. globigii spores in PBST (A) and E. coli O157:H7 in meat extract and PBST (B) using 3-mm-diameter PEG-modified glass beads. Standard errors of the means are masked by the symbols. The y axis indicates absorbance.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Standard curves of capture of B. globigii spores in environmental and industrial water samples using PEG-modified glass beads. Error bars represent standard errors of the means. The y axis indicates absorbance.

Stability of modified beads. Both dextran- and PEG-modified beads stored in the river and the laboratory lost activity within the first 2 weeks (Fig. 6). The pH and temperature of the river stayed between 8.23 and 8.38 and between 3 and 7°C, respectively. These results led us to repeat the experiment, taking a sample every day for 14 days. After 1 day, the ability of both river and laboratory beads to capture 10⁵ B. globigii spores declined. No further decrease was observed after day 3, and the beads showed similar capturing capabilities through 14 days. Differences in storage temperature (lab, 8°C; river, 1 to 4°C) and pHs (lab, 7.2; river, 8.1 to 8.5) did not seem to influence the stability of the beads.

View larger version (13K): [in this window] [in a new window]   FIG. 6.   Stability of anti-B. globigii modified glass beads using both dextran and PEG as spacers in river water and (A) the laboratory (B). Error bars represent standard errors of the means. The y axis indicates absorbance.

View larger version (27K): [in this window] [in a new window]   FIG. 7.   Effect of blocking buffers on signal generation using 103 total E. coli O157:H7 cells over time stored at 25°C () and 7°C (). Standard errors of the means are masked by the symbols.

Confocal microscopy. Hybridization slides modified with anti-B. globigii spore Ab and anti-E. coli O157:H7 Ab using PEG were examined with confocal microscopy. B. globigii spores were seen as bright blue oval shapes (Fig. 8), and live E. coli O157:H7 cells were seen as bright green spheres (Fig. 9). The slides captured both spores and cells. Control slides with no Ab attached did not show any capture. We used this technique to verify that the spores and cells were captured by the Ab-modified glass slides and not just physically adsorbed onto them.

View larger version (38K): [in this window] [in a new window]   FIG. 8.   Confocal image of captured B. globigii spores. Spores are seen as small blue spheres. Slides were coated with PEG and anti-B. globigii Ab. The inset graphically shows the capture of the spore (blue) by an immobilized antibody.

View larger version (33K): [in this window] [in a new window]   FIG. 9.   Confocal image of captured E. coli O157:H7 cells. Live cells are stained green, and dead cells are stained red. Slides were coated with PEG and anti-E. coli O157:H7 Ab.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The rapid detection of microbes in samples such as food is becoming more critical. Our research group has focused on developing a biosensor that is more rapid than and at least as sensitive as currently commercially available microbial tests kits. Currently commercially available kits for detecting bacteria in food rely on a preenrichment step that allows selective enrichment of the test organism, followed by rapid lateral flow ELISA or a PCR test (20, 21, 24). These tests vary in sensitivity but generally require ~10⁵ to 10⁸ CFU to generate a signal (unpublished data). This level is obtained during the preenrichment step, depending on the cellular state during incubation. Therefore, a need exists to develop a more sensitive test that produces results in a shorter time frame. The objective of this work was to eliminate the enrichment step to allow detection of bacteria directly in samples. Direct comparison of the ImmunoFlow test and commercial tests is currently being studied.

This new method described here, ImmunoFlow, involves the use of specific Ab immobilized onto glass beads (typically 3 mm in diameter) housed in a cartridge to allow flow in a fluidized bed format. The use of a fluidized bed allows real-time testing of samples that contain particulates commonly encountered in food and environmental samples and is independent of sample size. Detection of bound microbes involves the use of existing ELISA protocols.

Since this is an Ab-based method, it is directly dependent on the specificity of the immobilized Ab. Cross-reactivity of commercially available Ab was investigated to select the most species-specific Ab for immobilization. Using two different Ab as the 1° and 2° Ab in an ELISA format generally gives a greater signal than using the same antibody as both the 1° and 2° Ab (16). Therefore, we used each of the E. coli Ab mentioned in Table 1 as 1° Ab and tested them against the remaining three Ab. The combination of the polyclonal KPL Ab as 1° Ab with the monoclonal 3011 Ab as 2° Ab was very specific but gave a low signal. It was found that using KPL as both 1° and 2° Ab produced the strongest signal. Therefore, this Ab combination was used in the development of the ImmunoFlow system for the capture and detection of E. coli O157:H7. One of the problems encountered was the commercial availability of E. coli O157:H7 antibodies. A handful of Ab are available, most of which are monoclonal and thus very specific. The other antibody tested for cross-reactivity, anti-B. globigii Ab, showed no cross-reactivity to the spores tested; therefore, this antibody was used as both the 1° and 2° Ab. The specific capture of B. globigii and E. coli O157:H7 molecules was demonstrated with confocal microscopy (Fig. 8).

The capture of small molecules without flow was demonstrated with BSA and OVA (BSA molecular weight, 66,000; OVA molecular weight, 43,000) with PEG- or polyThr-modified beads (Fig. 1 and 2). BSA and OVA were used as the toxin simulants. Some of the toxins of concern to the food industry are staph and botulinum toxins and aflatoxins. Very small amounts of these toxins will cause the onset of disease; thus, more-sensitive detection methods need to be available. The kits available today will recognize small amounts of toxins, but the test time is 4 h or less. The test described in this paper detects <1 ng of small proteins using Dynalbeads, whereas 3-mm-diameter modified glass beads will detect up to 4 µg. These levels are sensitive enough to detect and quantify the toxin of interest. However, the test becomes inadequate at higher concentrations because BSA will bind nonspecifically to the beads in addition to specific antibody capture (32).

The capture of B. globigii spores from both buffer and river water was demonstrated using 7-mm-diameter ceramic beads by recycling solutions spiked with 10⁶ total spores at various flow rates (Fig. 3). There was variability in the spore capture efficiency with respect to the sample medium and flow rate. Some of this variation can be explained by the size of the sample that was collected. The amount of Ab per surface area is sufficient to capture 10⁶ spores. However, the randomly selected beads taken out as samples and tested for spore capture may or may not have spores attached to them. The optimum sample size would have included testing all the beads in the module and then returning them to the module for continued capture. However, this was not done for practical reasons, because once the test was performed on the beads they could no longer be used for capture.

A comparison of bead size and linker type in relation to the capture of B. globigii was demonstrated with ceramic beads with a polythreonine linker and 3-mm-diameter glass beads with a PEG linker (Fig. 3, 4A, and 5). Both bead types were capable of capturing B. globigii spores, with a linear relationship between the spore count and the signal generated (Fig. 4A). The lower limit of detection was 10 spores, and this limit can potentially be increased by adding more antibody-modified beads for capture. However, the upper limit with PBST was 10⁵ total spores, and for tank water it was 10³ total spores. All environmental and industrial samples had pHs ranging from 7.2 to 9.2. The beads were active and captured spores over this range. An unexpected result was found with tank water, because the pH was high, 9.2. At this pH, antibodies begin to denature but the bound antibodies were stable.

The signal drop-offs and plateaus observed within the standard curves (Fig. 4B and 5) can be explained. Overloading Ab-modified beads with cells or spores can create a "hook" effect commonly observed between two competing antibodies (35). Another explanation is the "falling off" (31) of spores or cells that were interacting with the antibodies on the beads. If too many targets are available for potential interaction, they compete for binding sites and thus interfere with one another. This may lead to dissociation between antibody and target. It may be necessary to use several dilutions of a sample to bring it within the detection range when using this test and a sample with an unknown number of cells or spores.

There was no increase in the signal when using slush tank or river water as samples. This may be due to the already high level of other, interfering, spores. This may also help to explain the high background level seen in the control. The protein present in the slush tank, which contained whey and other cheese production byproducts, can coat the surface of the modified beads and thus prohibit the interactions between the capture molecule and the target (32).

The capture of E. coli O157:H7 in flow with 3-mm-diameter glass beads containing a PEG spacer in two sample types was demonstrated (Fig. 4B). In both sample types, there was an increase in the signal with an increase in the number of cells, with a maximum signal generated at 10⁴ cells in meat extract and 10³ cells in PBST. These results are similar to what we observed with the capture of B. globigii, in that a true unknown sample may have to be tested at several dilutions to fit within the detection range of this test. There was a higher signal and background level observed in the meat extract sample. This phenomenon may be due to the interaction of the fat and proteins of the meat with the test components. Secondary and tertiary Ab are difficult to wash away completely, and thus nonspecific bindings of these antibodies contribute to the stronger signal generated.

The stability of the Ab beads and the influence of the spacer type were investigated with anti-B. globigii beads and PEG and dextran (Fig. 6.). The PEG beads consistently showed a higher signal and background level than the dextran beads, which is probably due to nonspecific interaction of the 2° and/or 3° Ab with the surface of the beads. Both bead types showed a decrease in activity beginning at day 1. This decrease continued to a plateau at 2 to 3 weeks, at which time the signal indicated that spore capture and detections were still possible under the conditions tested. Immobilized E. coli beads showed a similar decrease in activity from time zero to approximately 4 days (Fig. 7), independent of the blocker type and temperature of storage. The bead activity was then fairly constant from 4 to 8 days.

We then investigated the use of several blocking agents that can be used to reduce nonspecific adsorption of 2° and 3° Ab and maintain activity of the immobilized Ab beads (Fig. 7). We found that the stability of the beads was independent of the blocker type and temperature of storage. Guire (16) investigated the use of several blocking agents to reduce nonspecific binding. He blocked mono- and polyclonal Ab bound to 96-well plates with various blocking agents, dried the plates, and tested for activity over 18 months. Three blocking agents, Stabilguard, Stabilcoat, and Superblock, together with two controls, BSA and PBS, were used. All three blocking agents maintained 70 to 80% activity after 3 months, whereas PBS- and BSA-stored plates only maintained about 10% activity. In our lab the beads were stored in a BSA blocking liquid, and we observed an average 70% loss of activity after 4 days.

Conclusion. Various surfaces can successfully be modified with antibodies to capture small proteins, spores, and bacterial cells. This capture mechanism can easily be adapted for the capture of other targets, such as small toxins and other food-borne bacteria.